Application and Research on Fusion Model and XGBoost Based on Linear Regression and Gray Prediction

In this paper, a NEVC Dataset is constructed using 11 index data selected through principal component analysis and data transformation. Correlation analysis is performed based on Spearman correlation coefficients to test the significance of the metrics and build a ridge regression model for exploring the degree of influence of different factors. Forecasting is conducted using a fusion model of linear regression and gray prediction as well as a machine learning-based XGBoost model. Granger causality test is employed to analyze the impact of various influencing factors on the industry, draw conclusions, and predict future trends.